of aberration of some mutations in HCC [62] and of these, six
proteins were in the core network and five were agreed with model
prediction. A similar summary of the type of aberration of some
mutations in other cancer types [63] reveals that eight proteins
were in core network and seven were agreed with model prediction
(Table4). This overall agreement further supports the significant
potential of this analysis to predict genetic mutations in cancer.
However, it should be mentioned that there is one disagreement
between the model and the literature. It was well known that p53
has a loss-of-function mutation in many cancers, while p53 was
predicted to have a gain-of-function mutation in our model. This
disagreement may be owing to one of two sources. First, given the
heterogeneous nature of cancer mutations and the current incom-
plete network, experimental results of certain genes which do not
appear to fit this model are expected. Second, the aforementioned
results were obtained with a threshold as 0.4. We found that p53
was one of three genes whose results are sensitive to threshold
values. Thus, the behavior of p53 may be more complex than
presently believed, as some recent studies have suggested [64].
Our analysis also affords two additional intriguing and testable
predictions. First, our model suggests that there are mutations that
can confer selective advantages to establish and maintain the nor-
mal hepatocytes phenotype: a preferred mutation spectrum in the
normal hepatocyte (Table5). There is some evidence showing that
cells and tissues can maintain their normal phenotype in the face of
myriad mutated genes [65]. It should be biologically interesting to
determine whether there would be such a mutation spectrum in
normal liver. Second, while we have used this model to predict
mutations in cancer successfully as shown by the above analysis, it
showed that normal hepatocyte and cancerous hepatocyte are
endogenous stable states of one single endogenous network. This
indicated that there are cancers, especially at the early stage, which
can take place without major genetic alterations such as these well-
documented oncogenes and tumor suppressor genes. Indeed, there
is evidence supporting the existence of mutation-free cancer from
different standpoints [9, 66]. If firmly established, the cancer gen-
esis and procession is mechanistically completely different from that
of the cancer mutation theory.2.6 New Cancer
Therapy Strategies
from Endogenous
Network Model
and Theory
One of the most characteristic properties of dynamical systems is
each attractor is maintained by some key molecular-cellular agents
and their interactions. Positive feedback loops provide a simple
general strategy for the establishment and maintenance of heritable
phenotype [67]. The present working model reveals that the agents
and interactions maintain normal liver and HCC by forming dis-
tinct positive feedback loops in Fig.3 [68].
The model reveals that positive feedback loopHNF4α-C/EBP-
α-Foxa2 is responsible for the establishment and maintenance of232 Gaowei Wang et al.